---
title: Claude Code
description: Claude Code settings, hooks, plugins, slash commands, and skill links managed under ~/.claude.
---

Claude Code is Anthropic's terminal coding agent. The repo manages its user-level settings, hook scripts, custom slash commands, and the skill links it loads. The `claude` binary itself is not installed by chezmoi.

## Files

| Source in repo | Target | Notes |
| --- | --- | --- |
| `dot_claude/settings.json` | `~/.claude/settings.json` | Plain file; in symlink mode the target is a symlink back into the repo |
| `dot_claude/empty_CLAUDE.md` | `~/.claude/CLAUDE.md` | `empty_` keeps the zero-byte file; no global memory instructions |
| `dot_claude/hooks/executable_cmem-to-obsidian.py` | `~/.claude/hooks/cmem-to-obsidian.py` | Executable |
| `dot_claude/hooks/executable_herdr-agent-state.sh` | `~/.claude/hooks/herdr-agent-state.sh` | Executable, generated by herdr |
| `dot_claude/hooks/gitnexus/gitnexus-hook.cjs` | `~/.claude/hooks/gitnexus/gitnexus-hook.cjs` | Present but not wired in `settings.json` |
| `dot_claude/commands/*.md` | `~/.claude/commands/*.md` | User slash commands |
| `dot_claude/skills/symlink_*` | `~/.claude/skills/*` | Symlinks into `~/.agents/skills` |
| `dot_claude/skills/<dir>/` | `~/.claude/skills/<dir>/` | Claude-only skills kept as real directories |

None of these are role-gated. Hook commands in `settings.json` use absolute `/home/prad/...` paths; on a machine with a different home directory, edit them after cloning.

## settings.json

### Model and UI

| Key | Value | Effect |
| --- | --- | --- |
| `model` | `fable` | Default model alias |
| `tui` | `fullscreen` | Full-screen terminal UI |
| `agentPushNotifEnabled` | `true` | Push notifications when the agent needs attention |
| `skipDangerousModePermissionPrompt` | `true` | No confirmation when starting in bypass-permissions mode |
| `skipAutoPermissionPrompt` | `true` | No confirmation when entering auto mode |

No `statusLine` is configured; Claude Code uses its default status line.

### Environment

`env` sets two variables for every session:

| Variable | Purpose |
| --- | --- |
| `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Turns off Claude Code's built-in auto memory. Session memory comes from the claude-mem plugin instead. |
| `EDITOR` | Points the external editor at `micro`. |

### Permissions

- `defaultMode` is `auto`, so sessions start in auto mode and the classifier approves routine actions.
- The `allow` list pre-approves seven read-only tools from the `codegraph` MCP server: `codegraph_search`, `codegraph_context`, `codegraph_callers`, `codegraph_callees`, `codegraph_impact`, `codegraph_node`, and `codegraph_status`.
- There is no `deny` or `ask` list.

`autoMode.environment` is a list of trust-boundary notes that auto mode reads when judging risk. It declares no org, cloud provider, secrets manager, or CI target. It names one private personal LeetCode repository as the trusted repo and treats routine work there as low risk. It marks `.env*` files as sensitive. Any host, namespace, or container with `prod` or `production` as a whole name segment counts as a sensitive remote target. IaC changes touching IAM, RBAC, networking, quota, and node pools count as protected.

### Hooks

`settings.json` wires six hook commands:

| Command | What it does |
| --- | --- |
| `pixtuoid-hook` | External binary that receives lifecycle events. Matcher `.*`. |
| `~/.orca/agent-hooks/claude-hook.sh` | Orca's agent hook. Runs through `/bin/sh` only if the file exists and is executable, so machines without Orca skip it silently. |
| `atuin hook claude-code` | Records the agent's `Bash` tool calls in [atuin](/shell/atuin) history. Matcher `Bash`. |
| `python3 ~/.claude/hooks/cmem-to-obsidian.py` | Exports the session's claude-mem learnings to Obsidian ([below](#cmem-to-obsidian)). 15 s timeout. |
| `bash ~/.claude/hooks/herdr-agent-state.sh session` | Reports the session to herdr ([below](#herdr-agent-state)). 10 s timeout. |
| `sh ~/skills/tldraw-offline/inject-server-context.sh SubagentStart` | Injects server context when the `tldraw-offline` subagent starts. The script lives outside this repo. |

Event wiring:

| Event | pixtuoid | orca | atuin (`Bash`) | Other |
| --- | :-: | :-: | :-: | --- |
| `SessionStart` | yes | | | herdr-agent-state `session` |
| `UserPromptSubmit` | | yes | | |
| `PreToolUse` | yes | yes | yes | |
| `PermissionRequest` | | yes | | |
| `PostToolUse` | yes | yes | yes | |
| `PostToolUseFailure` | | yes | yes | |
| `Notification` | yes | | | |
| `SubagentStart` | yes | | | tldraw-offline context (matcher `tldraw-offline`) |
| `SubagentStop` | yes | | | |
| `Stop` | | yes | | |
| `StopFailure` | | yes | | |
| `SessionEnd` | yes | | | cmem-to-obsidian |

### Plugins

`enabledPlugins`:

| Plugin | Marketplace | State |
| --- | --- | --- |
| `caveman` | `caveman` | enabled |
| `claude-mem` | `thedotmack` | enabled |
| `context-mode` | `context-mode` | enabled |
| `context7` | `claude-plugins-official` | disabled |
| `council` | `council-of-high-intelligence` | enabled |
| `exa` | `claude-plugins-official` | enabled |
| `frontend-design` | `claude-plugins-official` | enabled |
| `gopls-lsp` | `claude-plugins-official` | enabled |
| `gopls` | `claude-code-lsps` | enabled |
| `lua-lsp` | `claude-plugins-official` | enabled |
| `ponytail` | `ponytail` | enabled |
| `solidity-language-server` | `claude-code-lsps` | enabled |
| `understand-anything` | `understand-anything` | enabled |
| `vtsls` | `claude-code-lsps` | enabled |
| `yaml-lsp` | `zircote-lsp` | enabled |
| `diagram-design` | `diagram-design` | enabled |

The LSP plugins give the agent language-server diagnostics for Go, Lua, Solidity, TypeScript/JavaScript (vtsls), and YAML.

### Marketplaces

`extraKnownMarketplaces` registers eleven plugin sources:

| Name | Source |
| --- | --- |
| `caveman` | GitHub `JuliusBrussee/caveman` |
| `claude-code-lsps` | GitHub `Piebald-AI/claude-code-lsps` |
| `claude-plugins-official` | GitHub `anthropics/claude-plugins-official` |
| `council-of-high-intelligence` | GitHub `0xNyk/council-of-high-intelligence` |
| `mempalace` | GitHub `MemPalace/mempalace` |
| `omc` | git `https://github.com/Yeachan-Heo/oh-my-claudecode.git` |
| `ponytail` | GitHub `DietrichGebert/ponytail` |
| `thedotmack` | GitHub `thedotmack/claude-mem` |
| `understand-anything` | GitHub `Egonex-AI/Understand-Anything` |
| `zircote-lsp` | GitHub `zircote/lsp-marketplace` |
| `diagram-design` | GitHub `cathrynlavery/diagram-design` |

`mempalace` and `omc` are registered but no plugin from them is enabled. `context-mode` is enabled but its marketplace is not declared here; it comes from Claude Code's own marketplace state.

## Hook scripts

### cmem-to-obsidian.py [#cmem-to-obsidian]

A stdlib-only Python script run on `SessionEnd`. It reads the hook JSON from stdin and exports that session's claude-mem observations to an Obsidian note. It fails soft: any error is logged and the script exits 0, so it never blocks the session.

1. Opens the claude-mem SQLite database read-only (`?mode=ro`), so it never locks the live DB or its WAL.
2. Maps the Claude Code `session_id` to claude-mem's memory session via `sdk_sessions.content_session_id`, then loads that session's observations and latest summary.
3. Drops any observation containing `<private>`.
4. Writes `<vault>/<project>/<date> <project> <session-prefix>.md`. The file name is keyed to the session, so re-runs overwrite the same note.
5. Creates `<vault>/<project>.md` as a map-of-content note if it is missing, so the `[[project]]` wikilink resolves.

Note layout: frontmatter (`date`, `project`, `session`, `source: claude-mem`, tags `claude-mem` and `project/<slug>`), an abstract callout from the session summary, then sections by observation type. Each bullet links back to the claude-mem viewer at `http://localhost:<port>/?observation=<id>`.

| Section | Observation types |
| --- | --- |
| Decisions | `decision` (rendered as note callouts) |
| Fixes & gotchas | `bugfix` |
| Insights | `discovery`, `feature`, `change`, `refactor`, plus any unmapped type |

| Variable | Default |
| --- | --- |
| `CLAUDE_MEM_DB` | `~/.claude-mem/claude-mem.db` |
| `CMEM_OBSIDIAN_DIR` | `~/Documents/Vault/90 - Archive/Learnings` |
| `CLAUDE_MEM_WORKER_PORT` | `37700` |

Log file: `~/.claude/hooks/cmem-to-obsidian.log`.

### herdr-agent-state.sh [#herdr-agent-state]

Generated and versioned by [herdr](/terminal/herdr) (integration id `claude`, version 7). The header says reinstalling herdr's integration overwrites it, so put custom hooks in separate files. It runs on `SessionStart` with the `session` action and exits immediately for any other action.

It only acts inside a herdr pane: `HERDR_ENV=1`, `HERDR_SOCKET_PATH`, and `HERDR_PANE_ID` must be set, and `python3` must be on `PATH`. It then sends one `pane.report_agent_session` request over herdr's Unix socket with the pane id, the Claude session id, the transcript path, and the session start source. herdr uses this to link the pane to the agent session. Subagent events and `SubagentStop` are ignored, so a finished subagent cannot mark an idle pane as working again. Socket errors are swallowed and the script has a 0.5 s timeout.

### gitnexus/gitnexus-hook.cjs

A Node `PreToolUse` handler that adds GitNexus code-graph context to searches. `settings.json` does not reference it, so it is inactive until you add a hook entry.

- Runs only when `.gitnexus` exists in the working directory or one of its four nearest ancestors.
- Pulls a search pattern from `Grep` (the pattern), `Glob` (the first identifier-like path segment), or `Bash` commands that call `rg` or `grep` (the first non-flag argument). Patterns under three characters are skipped.
- Runs `gitnexus augment <pattern>`. It tries the bun global install at `~/.bun/install/global/node_modules/gitnexus` first, then `require.resolve`, then `npx -y gitnexus` (8 s or 15 s timeout).
- Returns the CLI's stderr output as `hookSpecificOutput.additionalContext`.

## Slash commands

`dot_claude/commands/` becomes `~/.claude/commands/`, one user-level slash command per file.

| Command | Arguments | Purpose |
| --- | --- | --- |
| `/bd-drain` | optional priority filter (`p0-p2`), epic id, or `dry-run` | Works through every open beads issue until all are closed with tests and builds passing. Records a green baseline first; `dry-run` prints the dependency-sorted plan and stops. Allowed tools: `bd`, `git`, `go`, `pnpm`, `cargo`, `make`, file tools, `Task`. |
| `/cmem-to-beads` | none | Pulls insight, decision, and gotcha learnings out of claude-mem and saves them as beads memories with `bd remember`. Actionable items are skipped. |
| `/cmem-to-linear` | none | Turns actionable claude-mem learnings (bugs, tech debt, follow-ups) into beads tasks, then pushes the new ones to Linear. |
| `/create-task` | `<task description or goal>` | Breaks a task into beads issues with dependencies, acceptance criteria, and priorities. The prompt includes live `bd ready`, in-progress, and recently-closed lists. Allowed tools: `bd`, `git log`, `git status`, `rg`, `Task`. |
| `/rams` | optional file | "Rams" accessibility (WCAG 2.1) and visual design review of a component file. Offers a project scan when no file is given. |
| `/web-interface-guidelines` | `<file-or-pattern>` | Reviews UI code against the Vercel Web Interface Guidelines: accessibility, focus, forms, animation, typography, and more. |

## Skills

`~/.claude/skills/` holds two kinds of entries:

- **Symlinks** (`symlink_<name>` in the source). Each file contains a relative target such as `../../.agents/skills/blume`, so `~/.claude/skills/blume` resolves to `~/.agents/skills/blume`. The shared library is documented on the [Skills](/ai/skills) page.
- **Real directories** for Claude-only skills: `absurd`, `agents-sdk`, `cloudflare`, `cloudflare-email-service`, `cloudflare-one`, `cloudflare-one-migrations`, `create-beads-orchestration`, `durable-objects`, `sandbox-sdk`, `todoist-cli`, `turnstile-spin`, `web-perf`, `workers-best-practices`. See [Claude-only skills](/ai/skills#claude-only-skills).

## Repo-local settings

This repo's own `.claude/settings.json` applies only to Claude Code sessions opened inside the chezmoi source directory. chezmoi ignores dot-prefixed source entries, so it is never deployed. It runs `bd prime` on `SessionStart` and `PreCompact`, which reloads beads issue context at session start and before compaction.

## Integrations

- [kitty](/terminal/kitty) tab bar shows a robot glyph for tabs running `claude`.
- The fish `ai` picker launches `claude --dangerously-skip-permissions`; see [fish](/shell/fish).
- The claude-mem plugin feeds both `cmem-to-obsidian.py` and the `/cmem-to-*` commands.